Dissertation > Excellent graduate degree dissertation topics show

Research on Parallel Algorithm of Graph Sequence Community Mining Based on MapReduce

Author: TangJun
Tutor: ChenSongCan
School: Nanjing University of Aeronautics and Astronautics
Course: Applied Computer Technology
Keywords: graph sequence community structure cost function MapReduce parallelization
CLC: TP311.13
Type: Master's thesis
Year: 2012
Downloads: 73
Quote: 0
Read: Download Dissertation

Abstract


As an important branch of data mining, community mining of graph sequence is widelyapplied in practical problems such as social networks. It’s critical about how to obtain the valuableembedded information and how to accelerate algorithm on massive data sets. Current methodsalmost adopt the splitting algorithm based on tree-map record and bottom-up agglomerativealgorithm, and are static mining without consideration of time aspect. Based on coding cost, wepropose a graph sequence community mining algorithm GSCM, and propose GSCM-SC based onspectral clustering. With Hadoop Reduce parallel computation framework, we parallelizeGSCM-SC to propose the PGSCM parallel algorithm.Graph sequences in this paper are all binary. The paper firstly forwards the concept of graphsequence coding cost, and proposes the GSCM algorithm through optimizing this cost function.The algorithm makes balance between model complexity and community quality, so it candetermine the number of communities and their structures, not requiring any parameters. Initialstructure can be gained by clustering after information compression. Use the stochastic evolutionand selecting thought of genetic algorithm to avoid being trapped in local minima. It can timelydetect changes of community structure according to the coding cost fluctuation. And then validateGSCM via actual datasets.Based on spectral clustering, we propose GSCM-SC algorithm. By parallelizing performancebottlenecks, we propose parallel algorithm PGSCM. We parallelize the computing of similaritymatrix based on its independence of data points; parallelize the computation of Laplace matrixeigenvector using Lanczos method; parallelize K-Means based on independence of data points andcluster centers. And we construct a Hadoop system using several virtual machines to valid theeffectiveness of the algorithm and its performance promotion.At last, we introduce the gray aspect of graph sequence community mining, providing a goodresearch direction.

Related Dissertations

  1. The Changes of Microbial Community Along the Different Trophic Gradient of Seawater in Zhanjiang Bay,Q938.8
  2. Growth Pattern of Marsupenaeus Japonicus and Succesion of Plankton Community in Intensive Model,S968.22
  3. Arthropod Biodiversity in Paddy Fields under Different Management Regimes,S511
  4. Spatial Distribution Characteristics of Community Structure Components in Evergreen Broad-leaved Forest of China,S718.5
  5. Research on Algorithm of Parallel Copying Garbage Collection Based on Lisp 2 for Multicore System,TP332
  6. Studies on the Mesozooplankton and Microzooplankton Community in the Yellow River Estuary and its Adjacent Area in Summer and Autumn,Q958.8
  7. Mechanical System Reliability Optimal Allocation Based on Dynamic Programming and Software Development,TP311.52
  8. Mechanical System Reliability Allocation and Optimization and Software Development,TP311.52
  9. Optimal Control Design for Nonlinear Stochastic Systems with Input Delay,O232
  10. 3D Mannequins Generating Engine Based on eMTM with MapReduce,TP391.41
  11. Research of Main Memory Database Parallel Query Processing on SMP,TP311.13
  12. An Intrusion Detection System for High-Speed Networks,TP393.08
  13. The Research of Text Classification Based on Hadoop,TP391.1
  14. Research and Implementation on a Distributed Service Registry Based on HADOOP Platform,TP393.09
  15. Resarch of Task-level Data Processing Based on Multicore CPU and Test of Its Performance on Cluster Platform,TP274
  16. Hadoop data center deployment and tracking systems research,TP308
  17. Dynamic analysis based on multi-core acceleration method,TP332
  18. Fault Tolerance for MapReduce in the Cloud Environment,TP302.8
  19. High dynamic conditions SINS Moving Base Transfer Alignment Parallel Algorithms,V249.322
  20. Based on multi-core SMP cluster environment parallel ray tracing simulation of satellite imagery and Implementation,TP391.41
  21. Efficient multi- GPU based acoustic wave simulator and its application,TP391.41

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer software > Program design,software engineering > Programming > Database theory and systems
© 2012 www.DissertationTopic.Net  Mobile